03 March, 2021

QRPLabs 25 MHz TCXO performing well in the U3S

I have tried several oscillators in the U3S QRSS/WSPR transmitter from QRPLabs. First an analog devices DDS; then a Silicon Labs oscillator chip running from an ordinary 27 MHz crystal, and then QRPLabs own oven controlled oscillator. I thought I had managed to control the drift when I got the oven controlled oscillator. But then after a while it started drifting again. I didn't want to go through its rather cumbersome calibration procedure once more so I gave it up recently when the TCXO module became available. This tiny module is shown in the first image.

It turned out that it resolved all drift problems immediately and so I recommend it to all perfectionists like me who cannot stand to see a drift of 2, 3, or 4 Hz/min in the WSPR report. 

23 January, 2021

Eleven and 87 year sunspot cycles

It is well known that the sunspots vary over an eleven-year cycle. As the sunspot number increases, the ionosphere is more ionized and radio propagation on the high shortwave bands from about 10 - 30 MHz. Also, the higher the sunspot number, the more likely is the appearance of Northern Lights.

Here's spectral analysis of the monthly sunspot numbers from 1750-2020. First I show Fourier based spectrogram analysis. The y-axis is frequency or inverse years and there are arrows marking 11 years as well as the Gleissberg cycle of about 87 years. Note how the 11 year cycle has split into two periods since the 50's. There is a weak subharmonic at 5.5 years also. Due to the 90 year analysis window length, the time scale ends 45 years before 2020, i.e. in 1975. 

19 October, 2020

Longwave AM transmitter based on the XR-2206

Some years ago I described some AM transmitter experiments for the longwave band from 150 to 300 kHz. A transmitter that worked over a few meters was based on an XR-2206 function generator IC. Instead of building it from scratch as that first article indicated, I ended up purchasing and modifying the "Geekcreit® XR2206 Function Signal Generator DIY Kit Sine Triangle Square Output 1Hz-1MHz" from Banggood. 

That kit is not intended for AM modulation as the modulation input, pin 1, is grounded. It therefore needed these modifications:

03 September, 2020

GPS corrected 10 MHz reference

My  GPS corrected 10 MHz reference now plays beautifully, as I finally was able to continue the project after a long break. All the parts were here already in February, but projects don't always develop as planned. I ended up with a modification of the F2DC Versatile Low Cost GPS Corrected Frequency Standard. It consists of an Arduino Nano, a Si5351 I2C configurable clock generator, and a GPS where the 1 PPS pulse is used to correct the frequency every 40 seconds.

19 July, 2020

Tweaking the K2 for digital modes

My Elecraft K2 which I assembled as a kit in 2001, has mainly been a CW rig, but now I wanted to use it more for digital modes like FT8 also. That required some tweaking:
  1. Wider bandwidth in SSB filter, so I took it from 2.4 to 2.6 kHz (KI6WX)
  2. More sensitive VOX (G3RXQ)
  3. RF-filtering on audio input (G3RXQ)

All of these are modifications to the tighly packed KSB2 board shown above so it took some patience to implement them. The bandwidth modification consists in replacing six ceramic capacitors that were sitting between the crystals. The VOX sensitivity modification is done by adding a pot which can be seen in the image as it sits outside the PCB on the left-hand side. I was lucky to find a mini 5k slider pot in my junk box which fitted in. Finally the RF-filtering was done by adding an inductor of 2.5 mH (right below the 5k pot) and a decoupling capacitor.

06 July, 2020

Recognizing a CQ on the FT8 waterfall

A CQ is not so hard to recognize on an FT8 waterfall display as there is a charactaristic indendation from the right early in the 15 second period. This can be seen in the three circled traces in the figure.


They show a CQ at 711 Hz with SNR 11 dB, one at 899 Hz with SNR 16 dB where my receive cursor is placed, and one at 2003 Hz with SNR 6 dB. These are fairly strong signals so it is easy to see the indendation, but in my experience it works down to almost -10 dB also if the contrast is set correctly with the slider on the display. 

Look closely and you'll see one in the period before also, at 21:15:00 (SNR 12 dB). Try this the next time you run FT8 and I'm sure you'll recognize it too!


27 April, 2020

GPS Clock with Binary Display

Some years ago I started making a GPS Clock with a 20 by 4 LCD, Yet another Arduino clock. It has been running since then as my shack clock. Now was the time for some additional functions, in the form of fancy variations of how to show time. I wanted a display coded in 0's and 1's, but I found that there were more than one way to do that.

The simplest one to read may be BCD (binary-coded-decimal) displays. They can be horizontally oriented like here:


10 February, 2020

Si5351 GPS corrected VFO up and running

Arduino Nano, Si5351A clock generator, I2C LCD display, and rotary encoder. The GPS signal is connected to the plug to the lower left.
I got the Si5351A Arduino controlled GPS corrected VFO designed by W3PM, Gene Marcus, up and running the other day. It gets its timing information from a QRPLabs GPS and all the functions in the original article function as they should. The main change I made was to use an LCD display running over the I2C bus, thus saving a bunch of wires from the Arduino Nano to the display.

The software worked fine except that my locator was one off in the last letter, i.e. JO59fs rather than JO59fu. I hacked the original code by adding "1" to the variable GPSlocator[5] in the calcGridSquare routine. This is not a tested fix and it may possibly have other unknown and undesirable consequences, but it works in my location.

My plan is primarily to make a 10 MHz reference oscillator out of this, as a replacement for my "Just good enough 10 MHz GPS reference" which turned out to generate too much noise in the 14 MHz band.

Now on to decide what I want to show on the 2-line display for my use and then play with Arduino code. What I definitely want is some form of indication of correction factor and thus accuracy as in the version that F2DC built.


This blog post first appeared on the LA3ZA blog.